Related Experiment Videos
Using simulation data to predict lung geometry for inhomogeneity corrections in breast cancer treatments
1Department of Medical Physics, Rush-Presbyterian-St. Luke's Medical Center, Chicago, IL 60612, USA.
Purpose:
To develop a statistical model based only on simulation measurement data, to predict the lung geometry in the central slice of the tangential radiation treatment fields for breast cancer.
Methods And Materials:
A linear regression analysis was performed on 22 patients to determine the shape of lung in the central axis plane of the tangential radiation fields. Data collected include the greatest perpendicular distance (GPD) measured from the chest wall to the field border on computed tomography (CT) images, the central lung distance (CLD) measured from the posterior field border to the chest wall on the simulation portal images, and the lung contours digitized at 1 cm intervals. The lung contours of these patients were fitted to a parabolic curve through a polynomial regression model. A lung template based on the regression model is used to construct a "generic lung" contour on patients' external body contours for treatment planning. The accuracy of this technique was tested on another group of 15 patients for its ability to predict the shape of lung on the central axis plane and the accuracy of dose to the prescription point.
Results:
The polynomial regression indicates that all the patients' lung contours in the tangential fields follow a parabolic curve: Y = -0.0808 X2 + 0.0096 X + 0.0326. The maximum lung involvement (GPD) can be determined from the value of CLD measured on the simulation film by the linear regression model with a determination coefficient of 0.712. The 15-patient test results indicate that our model predicts the lung separation on the central axis with an average deviation of 1.35 cm, and the average absolute dose deviation to the dose prescription point is 1.46%.
Conclusion:
The model presented in this article provides an efficient method to estimate the lung geometry for breast cancer treatment planning without the requirement of CT data. The lung contour predicted by our model is useful for calculating dose distributions with inhomogeneity correction and may potentially benefit patients at higher risk of pulmonary toxicity.